Tools for Writing a Dissertation (and Studying)

Today, most of a researcher’s office work is conducted on a computer. Therefore, it is essential to have good, functional tools. In this article, I outline the research tools I currently use actively, and also mention software and workflows I have previously used but since abandoned.

With the advent of artificial intelligence, work methods are evolving rapidly, so note that I am writing this article during the Finnish July heatwave of 2025. These tips are intended for my seven-years-younger self starting a PhD, aiming to help avoid unnecessary pitfalls.

There are countless research tools across various fields for various uses, and for context, I completed my dissertation at Tampere University on asthma among cross-country skiers. The dataset involved about 700 participants, and the largest individual data file was roughly 50 megabytes—a modest size for modern computers. The raw data was gathered through an electronic questionnaire.

Hardware

I bought my first Apple computer at age 15 and have used them ever since. I also once owned the very first iPhone, which wasn't even sold in Finland initially. I have remained loyal to Apple's ecosystem.

Today, even the cheapest available MacBook (with Apple's M1, M2, etc., processors) is sufficient for writing and basic coding with smaller datasets. This is the recommendation I give to students I supervise. Additionally, I use a 13-inch iPad.

Software

For the final preparation of articles, I used Microsoft Word provided by my university’s subscription, and presentations were created with PowerPoint. Initially, I used Excel for manipulating raw data and categorizing open-text responses. As my skills advanced, I increasingly moved directly into coding, avoiding manual handling of raw data.

Choosing the right reference management system from the start is crucial. I previously used Refworks and Mendeley, until finally finding "The One"—Zotero. Zotero is open-source, allows seamless PDF storage, and keeps all literature in one place.

I've tried several note-taking applications, starting with Apple’s Notes app, then experimenting with Evernote, Notion, and briefly Microsoft's note-taking software during my studies.

However, I found myself returning to Apple's Notes app, which now holds over 2,000 notes spanning both professional and personal life. Practically every noteworthy item is stored there. I have a physical notebook, too, but mainly use it for jotting down NFL football scores and other sports results.

Figure 1: I’m not sure at which end of this intelligence-distribution Gaussian curve I fall in this regard, but I am convinced that a simple note-taking app makes work clearer. (Image source)

Coding

A doctoral candidate is likely to learn a new language—often a programming language. At Tampere Medical School, SPSS was taught during the second year, and we prepared seminar papers and assignments using it. I also used SPSS for my first two articles but found it cumbersome, particularly regarding high-quality visualizations.

After a few years with SPSS, I became enthusiastic about R, one of the most popular languages among scientists. With R, you can handle data, analyze it, and prepare publication-ready material from start to finish. The emergence of AI, particularly ChatGPT, drastically increased my coding productivity. I use R through RStudio, which is free for small-scale use.

I strongly recommend new researchers start learning R immediately, and I advocate teaching it as the primary language at medical schools. Although SPSS might initially be easier, R offers far greater versatility. However, I haven't yet explored Python.

Listening to experienced programmers has reinforced a recurring thought: Can you truly learn a language without writing it yourself?

Currently, my honest skill level includes:

• Knowing the steps needed to process small, locally stored datasets from raw data to final results.

• Understanding what packages, actions, and code are needed.

• Interpreting code written by ChatGPT and understanding each part.

• My biggest weakness: Struggling to identify and fix errors independently.

• Not knowing what I don't know—potentially better or more efficient methods than AI initially suggests.

Three practical tips for writing

Paid AI Tools

I've subscribed to ChatGPT for two years now, using a business plan (approximately €55/month as of July 2025). The biggest advantage comes from coding, with the subscription proving its value after just a few hours of coding sessions at my skill level.

AI is also helpful for English language refinement and proofreading, enhancing clarity and readability. However, I often remove half of its suggestions. A useful prompt for concise and clear text is: "Act as an academic assistant writer to a grumpy, blunt, and honest Finnish professor who knows not everything is certain."

I'm awaiting the day AI operates locally and securely enough for me to confidently upload research data. Despite ChatGPT's assurances of privacy for business users, we're not there yet. Currently, a few language models can run locally, but laptops lack sufficient power for effective use.

Research Diary

I have one massive note where I compile extensive notes with to-do lists, daily progress, and ongoing observations in.a chronological order. Generally, I never delete anything—not even scattered thoughts or incomplete observations.

Text to Speech (TTS)

Machine-generated voices have improved dramatically, with the best models sounding realistic and comfortable for extended listening. Currently, I mostly use Apple's built-in accessibility features, which read screen content aloud. Any book instantly becomes an audiobook without needing physical interaction—even automatically turning pages when reading e-books.

I've tried paid TTS apps like Speechify, featuring voices like Snoop Dogg and Barack Obama, but found it buggy, particularly with multi-column PDF articles. The price (~$12/month) led me to end my subscription after a year (tried 2023–2024).

Conclusion

When choosing computers and software, invest in tools that eliminate waiting time, leaving the technology to await your commands, not vice versa. I also appreciate easy-to-use software that allows quick initial use and has the depth for efficient, advanced functionality.

Enjoy your work!

How to Win in Ruka World Cup Sprint – The Battle of the Final Hill

A compilation video on one of the longest running sprint races in the Cross Country World Cup, the Ruka sprint. The video presents finals from 2009, offering insights into key moments and strategies where races are often decided on the final hill.

In the 17-year history of Ruka sprint and 34 finals, 24 of them could still be found on YouTube. Despite the track is short, it is one of the toughest: the last minute is everything.

150 m long final uphill with 14% incline leading to the stadium is a run for the money. Most of the races are decided in that hill and only sometimes the leader has changed in the final straight.

The tempo has changed massively through the years. The demands in strength and agility have increased speeds specifically in running up the hill and hammering that double poling in the final straight. The tactics have been always quite simple: have a good spot before the hill and then the strongest will (almost) always win.

All Winners in Ruka Sprint through the years:

2022-2023
Emma Ribom
Johannes Hoesflot Klaebo

2021-2022
Maja Dahlqvist
Alexander Terentev

2020-2021
Linn Svahn
Erik Valnes

2019-2020
Maiken Caspersen Falla
Johannes Hoesflot Klaebo

2018-2019
Yulia Belorukova
Alexander Bolshunov

2017-2018
Stina Nilsson
Johannes Hoesflot Klaebo

2016-2017
Stina Nilsson
Pal Golberg

2015-2016
Maiken Caspersen Falla
Sondre Turvoll Fossli

2014-2015
Marit Björgen
Eirik Brandsdal

2013-2014
Justyna Kowalczyk
Eirik Brandsdal

2012-2013
Marit Björgen
Nikita Kriukov

2011-2012 (videos missing)
Marit Björgen
Teodor Peterson

2010-2011
Marit Björgen
John Kristian Dahl (video missing)

2009-2010
Justyna Kowalczyk (video missing)
Ola Vigen Hattestad

2008-2009 (videos missing)
Petra Majdic
Ola Vigen Hattestad

2007-2008 (videos missing)
Petra Majdic
Johan Kjoelstad

2006-2007 (videos missing)
Petra Majdic
Jens Arne Svartedal